Download presentation
Presentation is loading. Please wait.
Published byPeter Harris Modified over 9 years ago
1
Spring 2015© CS 438 Staff - University of Illinois1 Next Topic: Vacation Planning UIUC Chicago Monterey San Francisco Chicago to San Francisco: ALL FLIGHTS FULL I’ll wait until Spring.
2
Spring 2015© CS 438 Staff - University of Illinois2 Congestion Avoidance Control vs. avoidance Control: minimize impact of congestion when it occurs Avoidance: avoid producing congestion In terms of operating point limits load power optimal load idealized power curve controlavoidance
3
Spring 2015© CS 438 Staff - University of Illinois3 Congestion Avoidance TCP’s strategy Control congestion once it happens Repeatedly increase load in an effort to find the point at which congestion occurs, then back off Alternative Strategy Predict when congestion is about to happen and reduce the rate at which hosts send data just before packets start being discarded Congestion avoidance, as compared to congestion control Two possibilities Host-centric TCP Vegas (may get some help from routers as in DECbit or via RED gateways) Router-centric Virtual circuits with reserved resources (ATM, RSVP)
4
Spring 2015© CS 438 Staff - University of Illinois4 DECbit (Destination Experiencing Congestion Bit) Developed for the Digital Network Architecture Basic idea One bit allocated in packet header Any router experiencing congestion sets bit Destination returns bit to source Source adjusts rate based on bits Note that responsibility is shared Routers identify congestion Hosts act to avoid congestion
5
Spring 2015© CS 438 Staff - University of Illinois5 DECbit Router Monitors length over last busy + idle cycle Sets congestion bit if average queue length is greater then 1 when packet arrives Attempts to balance throughput against delay smaller values result in more idle time larger values result in more queueing delay Queue length Current time Time Current cyclePrevious cycle Averaging interval
6
Spring 2015© CS 438 Staff - University of Illinois6 DECbit End Hosts Destination echoes congestion bit back to source Source records how many packets resulted in set bit If less than 50% of last window had bit set Increase CongestionWindow by 1 packet If 50% or more of last window had bit set Decrease CongestionWindow by 0.875 percent Note: Techniques used in DECbit known as explicit congestion notification (ECN) Proposal to add ECN bit to TCP in progress
7
Spring 2015© CS 438 Staff - University of Illinois7 Router-Based Congestion Avoidance Random Early Detection (RED) gateways Developed for use with TCP Basic idea Implicit rather than explicit notification When a router is “almost” congested Drop packets randomly Responsibility is again shared Router identifies, host acts Relies on TCP’s response to dropped packets
8
RED Overview Observation Transient congestion Should be accommodated for by having large enough queues Longer-lived congestion Reflected as an increase in the average queue size Approach Detect incipient congestion from average queue size Upper bound for average queue length Spring 2015© CS 438 Staff - University of Illinois8
9
RED Overview Notify the end host of congestion Dropping packet Marking packet Select connections randomly Avoid global synchronization Change dropping probability dynamically Avoid bias against bursty data Use average queue length Random marking Spring 2015© CS 438 Staff - University of Illinois9
10
Spring 2015© CS 438 Staff - University of Illinois10 Random Early Detection (RED) Hosts Implement TCP congestion control Back off when a packet is dropped Routers Compute average queue length (exponential moving average) AvgLen = (1 – Weight) * AvgLen + Weight * SampleLen 0 < Weight < 1 (usually 0.002) SampleLen is queue length at packet arrival time
11
Spring 2015© CS 438 Staff - University of Illinois11 RED – Dropping Policy If AvgLen MinThreshold Enqueue packet If MinThreshold < AvgLen < MaxThreshold Calculate P and drop arriving packet with probability P If MaxThreshold AvgLen Drop arriving packet MaxThresholdMinThreshold AvgLen
12
Spring 2015© CS 438 Staff - University of Illinois12 RED – Dropping Probability Computing P P is a function of AvgLen and Count Count is the number of packets that have arrived since last reset Reset happens when either a packet is dropped or AvgLen is above MaxThreshold TempP = (MaxP) * (AvgLen – MinThreshold) MaxThreshold - MinThreshold P = TempP (1 – count * TempP) AvgLen P(drop) 1.0 MaxP MinThreshMaxThresh
13
Calculate Average Queue Size Low pass filter If idle: Example: w q = 0.002 Spring 2015© CS 438 Staff - University of Illinois13 max th = 15 min th = 5 instantaneous queue size average queue size
14
min th and max th Determined by the desired average queue size Should be set sufficiently to maximize network utilization min th Controls the size of bursts max th Depends on the maximum average delay max th - min th Should be larger than increase in average queue size in one round trip time Avoid global synchronization Spring 2015© CS 438 Staff - University of Illinois14
15
Spring 2015© CS 438 Staff - University of Illinois15 RED Parameters MaxP is typically set to 0.02 When the average queue size is halfway between the two thresholds, the gateway drops roughly 1 out of 50 packets. MinThreshold is typically max/2 Choosing parameters Carefully tuned to maximize power function Confirmed through simulation But answer depends on accuracy of traffic model
16
Spring 2015© CS 438 Staff - University of Illinois16 Tuning RED Probability of dropping a particular flow’s packet(s) Roughly proportional to the that flow’s current share of the bandwidth If traffic is bursty MinThreshold should be sufficiently large to allow link utilization to be maintained at an acceptably high level If no buffer space is available, RED uses Tail Drop Difference between two thresholds Should be larger than the typical increase in the calculated average queue length in one RTT Setting MaxThreshold to twice MinThreshold is reasonable for traffic on today’s Internet Penalty Box for Offenders
17
Spring 2015© CS 438 Staff - University of Illinois17 Source-Based Congestion Avoidance Idea Source watches for some sign that some router’s queue is building up and congestion will happen soon Examples RTT is growing Sending rate flattens
18
Spring 2015© CS 438 Staff - University of Illinois18 Source-Based Congestion Avoidance Observe RTT If current RTT is greater than average of minRTT and maxRTT, decrease congestion window by one-eigth Observe RTT and Window Size Adjust window once every two RTT If (CurrWindow – OldWindow) * (CurrRTT – OldRTT) > 0, decrease window by one-eigth, otherwise increase window my one MSS Observe sending rate Increase window and compare throughput to previous value Observe throughput Compare measured throughput with observed throughput TCP Vegas
19
Spring 2015© CS 438 Staff - University of Illinois19 TCP Vegas Time (seconds) 900 300 100 0.51.01.54.04.56.58.0 Observed Throughput KBps 1100 500 700 2.02.53.03.55.05.56.07.07.58.5 Time (seconds) 0.51.01.54.04.56.58.0 Queue size in router 5 10 2.02.53.03.55.05.56.07.07.58.5
20
Spring 2015© CS 438 Staff - University of Illinois20 TCP Vegas Basic idea Watch for signs of queue growth In particular, difference between increasing congestion window stable throughput (presumably at capacity) Keep just enough “extra data” in the network Time to react if bandwidth decreases Data available if bandwidth increases
21
Spring 2015© CS 438 Staff - University of Illinois21 TCP Vegas Implementation Estimate uncongested RTT baseRTT = minimum measured RTT Expected throughput = congestion window / baseRTT Measure throughput each RTT Mark time of sending distinguished packet Calculate data sent between send time and receipt of ACK
22
Spring 2015© CS 438 Staff - University of Illinois22 TCP Vegas Act to keep the difference between estimated and actual throughput in a specified range Below minimum threshold Increase congestion window Above maximum threshold Decrease congestion window Additive decrease used only to avoid congestion Want between 1 and 3 packets of extra data (used to pick min/max thresholds)
23
Spring 2015© CS 438 Staff - University of Illinois23 TCP Vegas Algorithm Let BaseRTT be minimum of all measured RTTs Commonly the RTT of the first packet If not overflowing the connection, then ExpectRate = CongestionWindow/BaseRTT Source calculates sending rate (ActualRate) once per RTT Source compares ActualRate with ExpectRate Diff = ExpectedRate – ActualRate if Diff < a Increase CongestionWindow linearly else if Diff > b Decrease CongestionWindow linearly else Leave CongestionWindow unchanged
24
Spring 2015© CS 438 Staff - University of Illinois24 TCP Vegas Algorithm Parameters = 1 packet = 3 packets Even faster retransmit Keep fine- grained timestamps for each packet Check for timeout on first duplicate ACK 0.51.01.52.02.53.03.54.04.55.05.56.06.57.07.58.0 CAM KBps 240 200 160 120 80 40 Time (seconds)
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.